Spot on Bigun!!! You are all adults, I expect you to get over it and put up or shut up. If you throw a stone don't be surprised if one comes back at you.
There are many 'Blameless' amps around. Would you like to link to one which you would like us to equal or beat with a similar or simpler CFA? Please list the performance parameters you would like to consider.forr said:
Which CFAs have you you built that requires this level of matching?
And BTW, which VFAs have you built that require serious matching in the differential stages too?
kgrlee said:
For reality, you may like to consider bonsai's excellent designs that demonstrate general good performance with simplicity.forr said:
But which specific myth are you referring too? If you pluck a myth out of thin air, we may not be able to give you a concrete example immediately but should be able to show a realistic example in SPICE world within a reasonable time scale.
If you provide us with the specific Blameless example as above, we can use it as the benchmark for a bit of myth-busting.
Last edited:
An example of a classic voltage feedback amp would be Doug Selfs blameless which in outline covers the majority of amps available. Typically it has the common LTP (long tailed pair) input stage.
There are examples of "current feedback" designs within this thread... I wouldn't want to single one out as an example.
The JLH69 and similar designs such as my FET amp can also be referred to as "current feedback". Typically these use a "singleton" or single ended input stage.
Its worth searching out information on current feedback opamps as these typify the CFB architecture.
Not necessarily a singleton. VSSA, SX, NX among others use a complementary input stage.
Interestingly, JLH never referred to his amplifier as a so-called "CFA"; this, obviously, is because it is no such thing: it is a voltage feedback amplifier, although many ill-informed people call it a so-called "CFA".
See:
~ Scanned and cleaned up Wireless World Articles ~
Moreover, even those amplifiers with a complementary common-emitter input stage are also voltage feedback amplifiers, since the feedback quantity sampled in parallel with the output is a voltage, and remains a voltage because it is then applied in series with the input voltage; therefore so-called "CFAs" are a complete misnomer.
Last edited:
Just an observation... not a criticism --- one just needs to read the beginning where I made a determined attempt for the record to give literature, diagrams and formulas - all of which describe what a CFA is. And, Scott Wurcer comments as well. Its all there - just go back and read it for the late-comers who seem to think we haven't covered this yet.
There you will see exactly why the JLH is not a CFA. It is a VFA. And, as was also mentioned that there is now -over the decades- a common or street like slang use which says any circuit that has its feedback going to a low impedance point is a CFA. Which is were the mistake has been allowed to occure and where the confusion comes in. Though industry is more careful and knows the difference and makes both and sometimes combination circuits using VFB and CFB.
Thx-RNMarsh
There you will see exactly why the JLH is not a CFA. It is a VFA. And, as was also mentioned that there is now -over the decades- a common or street like slang use which says any circuit that has its feedback going to a low impedance point is a CFA. Which is were the mistake has been allowed to occure and where the confusion comes in. Though industry is more careful and knows the difference and makes both and sometimes combination circuits using VFB and CFB.
Thx-RNMarsh
Last edited:
The race to simplicity
It has been argued that for similar output stages, CFAs have less active devices than VFAs and are simpler.
Let's count them for two comparable configurations.
CFA
diamond input stage : 4 bipolar transistors
buffer of the VAS : 2 bipolar transistors
VAS : 2 bipolar transistors
Sum : 8 bipolar transistors in all.
VFA
We may consider Constant Current Sources as high values resistors (in fact they are high output impedance devices) and being passive rather than active devices (this may be a little specious but see further). Then :
differential + current mirror input stage : 4 bipolar transistors
input buffer of the VAS : 1 bipolar transistor
VAS : 1 bipolar transistor
Sum : 6 bipolar transistors in all.
If the CCS are considered are active devices, we add 3 bipolar transistors (one for the voltage reference). The number of bipolar transistors is now 9 in all, just one more than for the VFA above.
However, with a circuit using CCSs, the quiescent currents through the active transistors which determine their gm are settled independantly of their own characteristics and we do not have to check them.
This undoubtedly simplifies the controls once the circuits are mounted.
It has been argued that for similar output stages, CFAs have less active devices than VFAs and are simpler.
Let's count them for two comparable configurations.
CFA
diamond input stage : 4 bipolar transistors
buffer of the VAS : 2 bipolar transistors
VAS : 2 bipolar transistors
Sum : 8 bipolar transistors in all.
VFA
We may consider Constant Current Sources as high values resistors (in fact they are high output impedance devices) and being passive rather than active devices (this may be a little specious but see further). Then :
differential + current mirror input stage : 4 bipolar transistors
input buffer of the VAS : 1 bipolar transistor
VAS : 1 bipolar transistor
Sum : 6 bipolar transistors in all.
If the CCS are considered are active devices, we add 3 bipolar transistors (one for the voltage reference). The number of bipolar transistors is now 9 in all, just one more than for the VFA above.
However, with a circuit using CCSs, the quiescent currents through the active transistors which determine their gm are settled independantly of their own characteristics and we do not have to check them.
This undoubtedly simplifies the controls once the circuits are mounted.
First -- Welcome back everyone !!! 🙂
We have already done this. the VFA doesn't need a lot said here that isn't already available in current books by Cordell or Self. So we concentrated on the CFA which there are no equivalent books on... except in IEEE and other tech source papers and literature -- some references were given earlier.
It wasn't clear to most here and lacking books on the subject - so we are trying to understand it ourselves. This is where SIM's have helped a lot.
Thx-RNMarsh
Last edited:
One of the areas which seems in my experience that is easier in practice is the compensation or stabilizing the circuit and the interaction or trade-off with reducing distortion with VFA. I think this area will show some unique differences... particulary in practice. They have in my designs and I have made both types. Comparisons and greater understanding in this area are greatly appreciated.
Thx-RNMarsh
Thx-RNMarsh
Last edited:
I stand corrected then 🙂 I'll admit I've never made a study of CFA amps in the past and as such just "accepted" the terminology used as and when I came across it.
Thanks 🙂
If the figure at the left of the joined image is called a CFA input, there is no reason why the figure at the right should not be called a CFA input as well.
The input emitter follower of this figure at right is just here to bias the transistor receiving the feedback signal (I won't say if it's current or voltage), it can be replaced by another scheme - Linsley-Hood has shown one in a phone amp - and I think it still merits the now (more or less) admitted name of CFA.
Attachments
forr, why don't you tell us which Blameless version is your examplar? One you've built is good cos we have a 'real life' benchmark.
Also which VFAs & CFAs you've built where you've found the need for careful selection of dem devices.
We can then propose a simple mythbusting CFA. You can take pot shots at that and others can count active devices etc & decide if it is simpler.
Hi Forr,
Actually, the figures you've posted show that the feedback signal is applied in series with the input signal in both cases.
This means that the feedback signal at the point of application is a voltage and not a current, because if it were a current it would be applied in parallel (shunt-applied feedback) with the input signal.
Therefore, to refer to either scheme as a "current feedback" arrangement is really rather silly, like kgrlee.
Further reading:
http://www.diyaudio.com/forums/soli...current-feedback-amplifier-8.html#post3414948
Last edited:
Richard,
It seemed that we were moving right along in this thread at one point and then, blam, we are stuck back into semantic arguments again. I was looking forward to the answer of which topology would solve the main question of what is the best performing output section, a triple or whatever to complete the puzzle? I haven't seen much in the way of solutions to that major problem which most of you Gurus have identified as the major limitation of the design. I am still watching and waiting for some intelligent discussions working out that section for a true high powered amplifier.
ps. Still have those enclosures sitting waiting for you, I still want to see what your feelings are about those speakers we traded!
It seemed that we were moving right along in this thread at one point and then, blam, we are stuck back into semantic arguments again. I was looking forward to the answer of which topology would solve the main question of what is the best performing output section, a triple or whatever to complete the puzzle? I haven't seen much in the way of solutions to that major problem which most of you Gurus have identified as the major limitation of the design. I am still watching and waiting for some intelligent discussions working out that section for a true high powered amplifier.
ps. Still have those enclosures sitting waiting for you, I still want to see what your feelings are about those speakers we traded!
This sentence seems to sum up the issues, and seems more important than what the strict definition of CFA is.
That having been said, we need to be mindful of our definitions of simplicity and performance.
Is the goal of simplicity to save money or to satisfy the assertion by some that a simpler signal path must sound better?
Is the performance objectively measured or simulated performance, which can be objectively discussed, or does performance here include discussions of how it sounds independent of the measurements. Although how it sounds is ultimately what is most important, if you let that camel into the tent in this discussion, the discussion may become intractable.
With regard to simplicity, let's be mindful that transistors and resistors are incredibly cheap, but that adjustment pots and transistor matching are incredibly "expensive". If you are discussing a "CFA" that has only one pot (the output stage bias setting) and needs no more transistor matching than a VFA, now you are talking.
One man's fair apples-apples comparison may be another man's unfair apples-oranges comparison.
What if high slew rate is achieved, but only because the input stage can go into class B? I say it doesn't count because I don't think any stage in an amplifier, other than the output, should need to be other than class A. But that is an opinion on my part, and many may disagree.
Cheers,
Bob